Muon Spectrometer Phase-I Upgrade for the ATLAS Experiment: the New Small Wheel Project

CIPANP2018-Lefebvre October 1st, 2018

Muon Spectrometer Phase-I Upgrade for the ATLAS Experiment: the New Small Wheel project

Benoit Lefebvre on behalf of the ATLAS Muon Collaboration

Department of Physics McGill University, 3600 rue University, Montr´eal(Qu´ebec), H3A 2T8, CANADA

The instantaneous luminosity of the Large Hadron Collider at CERN will be increased by up to a factor of five to seven with respect to the design value. To maintain an excellent detection and background rejection capability in the forward region of the ATLAS detector, part of the muon detection system will be upgraded during LHC shutdown periods with the replacement of part of the present first station in the forward regions with the so-called New Small Wheels (NSWs). The NSWs will have a diameter of approximately 10 m and will be made of two detector technologies: Micromegas and small-strip Thin Gap Chambers (sTGC). The physics motivation for this significant upgrade to the ATLAS detector will be presented. The design choices made to address the physics needs will be discussed. Finally, the status of the production of the detector modules will be presented.

PRESENTED AT arXiv:1810.01394v1 [physics.ins-det] 2 Oct 2018 Conference on the Intersections of Particle and Nuclear Physics Palm Springs, USA, May 29 – June 3, 2018

c 2018 CERN for the beneﬁt of the ATLAS Collaboration. Reproduction of this article or parts of it is allowed as speciﬁed in the CC-BY-4.0 license. 1 Introduction

The Large Hadron Collider (LHC) [1] is a particle accelerator located at the CERN 34 2 1 laboratory. The initial design luminosity of the LHC is 1 10 cm− s− with a center-of-mass energy of 14 TeV during proton-proton collisions.× The luminosity has been gradually increased since the start of LHC operations and reached the design value in 2016. A series of upgrades are planned over the next decade during Long Shutdown (LS) periods to further increase the luminosity. In particular, the LHC will be upgraded to its High-Luminosity configuration (HL-LHC) during LS3, start- ing in 2024, which will bring the instantaneous luminosity up to 5 to 7 times the 1 design value [2]. The LHC is expected to deliver 3000 fb− of collision data by its decommissioning in 2037. An increased LHC luminosity brings more opportunities for physics discoveries to ATLAS [3], one of the particle physics experiments of the LHC. The additional amount of collision data collected will allow for more precise Standard Model measurements and an increased sensitivity to new physics processes. Data taking at high luminosity is, however, a challenge for ATLAS. The planned increase in luminosity will result in increased readout rates and particle fluences on the ATLAS detector systems. In order to fully benefit from the physics opportunities of the upgraded LHC, the Phase-I and Phase-II [4] upgrades of ATLAS will be carried out during LHC shutdown periods. In particular, as part of the Phase-I upgrade, the ATLAS muon identification ability in the forward regions of the detector will be improved. This upgrade project consists of replacing part of the ATLAS muon spectrometer by arrangements of muon detector modules called New Small Wheels [5] (NSWs) which are targeted for installation during LS2. This article is arranged as follows. A description of the ATLAS muon spectrometer is given in Section 2. Section 3 presents the physics motivation behind the NSWs installation. A description of the NSW design is given in Section 4. A status update of detector construction is presented in Section 5. A short summary of the article is given in Section 6.

2 ATLAS and the muon spectrometer

ATLAS is a multi-purpose detector of cylindrical geometry located at one of the LHC interaction points. The ATLAS detector systems are, in order of distance from the interaction point, the inner detector, the calorimeters and the muon spectrometer. ATLAS is also equipped with magnet systems which include the end-cap toroid magnets and the barrel toroid magnet. The magnetic ﬁeld generated by the magnet systems bends charged particles such as muons. In the end-cap regions, muons are

1 typically bent in planes of constant φ1. The ATLAS trigger and data acquisition system (TDAQ) controls the flow of data between the detector systems and permanent data storage. The TDAQ system follows a multi-level architecture. The first level, called Level-1, is hardware-based meaning it uses FPGAs2 and custom electronics to quickly process detector hits. The muon spectrometer is the ATLAS detector system responsible for muon measurements. The arrangement of muon detector modules follows an approximate 8-fold azimuthal symmetry. The barrel region has 3 cylindrical shells of detector modules, or stations. The end-cap regions have 3 disk-shaped stations. The innermost end-cap station is located at z = 7.4 m. The detectors making± up the muon spectrometer are gaseous ionization chambers. Different detector technologies are used for either Level-1 triggering or for precision measurements. Thin Gap Chambers (TGC) and Resistive Plate Chambers (RPC) are used for triggering. Trigger chambers are characterized by a fast response compared to precision chambers. Hits from trigger chambers are transmitted to the Level-1 trigger processor. A Level-1 trigger is issued if one or more muons in the appropriate transverse momentum pT range are identified based on hits from the trigger chambers. Monitored Drift Tubes (MDT) and Cathode Strip Chambers (CSC) have an excellent spatial resolution and are used for precision muon track space point measurements. Space points from the precision chambers are used for the measurement of the muon momentum based on the Lorentz curvature of the muon tracks in the magnetic field.

3 Motivation for the New Small Wheel Upgrade

The ATLAS Level-1 trigger rate increases linearly as a function of the LHC instantaneous luminosity [6]. The Level-1 trigger rate exceeds the ATLAS TDAQ data bandwidth when extrapolating to the luminosity expected during Run 33. Approxi- mately 80% of Level-1 triggers are associated with single muon candidates from the end-cap regions [5]. As shown in Fig. 1, 90% of single muon Level-1 triggers originate from reconstructed track segments that cannot be matched oﬄine with muons coming from the interaction point. A large fraction of the spurious Level-1 triggers originates from muon candidates observed in the end-cap regions. The fake muon rate is explained by

1ATLAS uses a right-handed coordinate system with its origin at the nominal interaction point in the center of the detector and the z-axis along the beam direction. The x-axis points from the interaction point to the center of the LHC ring; the y-axis points upward. Cylindrical coordinates (r, φ) are used in the transverse plane, where φ is the azimuthal angle around the z-axis. The pseudorapidity is deﬁned as η = ln(tan(θ/2)), where θ is the polar angle. 2Field-Programmable Gate Array 3The ATLAS Level-1 trigger data bandwidth is currently 100 kHz and will be 1 MHz after the ATLAS Phase-II upgrade.

2 background radiation incident on the end-cap middle station that is incorrectly iden- tiﬁed as muons by the trigger processor. The background radiation consists mostly of neutrons and photons that are generated in the material between the inner and the middle stations where the end-cap toroid magnets are located. Events

Figure 1: Pseudorapidity distribution of muon candidates identified at Level-1 passing Figure 2.6: ⌘ distribution of Level-1 muon signal (pT > 10 GeV) (L1_MU11) with the distribution of p the 11 GeVthe subsetT threshold with matched (L1 MU11 muon) and candidate of L1 MU11 (withinmuonR< candidates0.2)toano matchingffline well a reconstructed muon reconstructedmuon (combined by an offline inner algorithm, detector and with muon and spectrometer without a requirement track with p onT > its3 GeV), measured and offline pT [5]. reconstructed muons with pT > 10 GeV.

A possible workaround for the excessive trigger rate consists of raising the trigger p threshold from 20 GeV, the nominal value, to 40 GeV or in applying trigger MeasureT the second coordinate with a resolution of 1–2 mm to facilitate good linking between •prescaling. Both options would successfully reduce the Level-1 trigger rate to an the MS and the ID track for the combined muon reconstruction.. acceptable level, but at the expense of reducing the sensitivity to many physics pro- Thecesses. background In particular, environment a degradation in which ofthe the NSW muon will trigger be operating efficiency will at cause low p theT would rejection of manybe hits detrimental as spurious, for asHiggs they studies might because have been muons caused are by an important-rays, neutron signature or other for backgroundmany particles.processes Furthermore involving in the the Higgs life-time boson. of the detector, detection planes may fail to operate properly, with veryApart limited from opportunities issues arising for from repairing the excessive them. Hence Level-1 a multi-plane trigger rate, detector a performance is required. Anydegradation new detector of muon that might inner be end-cap installed station in the detectors place of theis expected current Small due to Wheel the high should be 2 1 operationalparticle for fluences. the full life A particletime of ATLAS flux reaching (and be up able to to 15 integrate kHz/cm 3000is expectedfb ). Assuming close to al least 10 yearstheof beam operation pipe during and the HL-LHC above expected operations. hit rate At per that second, level of approximately particle flux,10 the12 hits/cm CSC 2 are expectedand MDT in total chambers in the hottest will suffer region from of the unacceptable detector. inefficiencies4. The loss of muon precision space points will deteriorate the muon momentum resolution and increase the systematic error of physics studies. 2.3 TriggerThe proposed selection solution for reducing the trigger rate is to improve the online muon Performanceidentification studies in using order collision to recognize data and have reject shown fake the muons. presence A of schematic unexpectedly diagram high of rates of fake triggers4Inefficiencies in the end-capare already region. sizable Figure but reasonable 2.6 shows at the the nominal⌘ distribution LHC luminosity. of candidates selected by the ATLAS Level-1 trigger as muons with at least 10 GeV. The distribution of those candidates that indeed have an offline reconstructed muon track is also shown, together with the muons 3 reconstructed with pT > 10 GeV. More than 80% of the muon trigger rate is from the end-caps ( ⌘ > 1.0), and most of the triggered objects are not reconstructible offline. | | Trigger simulations show that selecting muons with pT > 20 GeV at Level-1 (L1_MU20) one would get a trigger rate at ps=14TeV and at an instantaneous luminosity of 3 1034cm 2s 1 of ⇥ approximately 60 kHz, to be compared to the total available Level-1 rate of 100 kHz. In order to estimate the effect of a trigger using the NSW a study has been performed applying offline cuts to the current SW to reduce the trigger rate. Table 2.1 shows the relative rate of L1_MU20 triggers in the ⌘ > 1.3 region after successive offline cuts to select high quality muon | | tracks. The various successive cuts applied are: i) the presence of Small Wheel track segments

18 the improved trigger algorithm for muon identification is shown in Fig. 2(a) where tracks A, B and C are identified as muons by the middle station. In the new trigger scheme, tracks are identified as muons onlyNew if they small feature wheel hits in both the inner and middle stations. In addition, the candidateEM muon track segment reconstructed by the inner station(a) must point to the interaction point and match the middle station measurements. Fake muons (track candidates BKill and the C) fake typically trigger do not have these features and would be rejected by the algorithm.by requiring IP pointing segment The implementation ofEI this enhanced trigger algorithm necessitates inner station C in EI (small wheel) detectors with excellent online track reconstructionA abilities with stable performances up to the expected level of particle fluencesB duringEquipped HL-LHC with operations. The current inner end-cap station detector cannot fullfil theseprecision requirements. tracker Therefore, part of the inner end-cap station will be replaced by thethat New works Small up Wheelsto (NSWs), shown in Fig. 2(b) and described in the next section. the ultimate luminosity, 5-7x1034 , with some EM (b) safety margin

fake 1 Trigger rate reduction muon studied using 0.8 1.3<|"|<2.5 EI data C s= 7 TeV 0.6 A Data 2011

B Reduction for L1_MU20 0.4 fake ~ 1/6 muon 0.2 in 1.3<η<2.5

26.03.2012 inner middle T.&Kawamoto0 interaction All L1_MU20 Inner_Seg>0 d! cut dL cut Offline Pt>20 point (IP) end-cap muon stations

(a) (b)

Figure 2: (a) Schematic diagram of the proposedUpdate trigger algorithm on Cathode used for fakeBoards muon production @ Triangle discrimination viewed on a quadrant cross-sectionSergey of Issinski ATLAS(TRIUMF) in the r-z plane. (b) Cut-away view of a NSW. Oliver Stelzer-Chilton (TRIUMF),

4 The New Small Wheels

The New Small Wheels (NSWs) are disk-shaped arrangements of 8 large and 8 small pie-slice detector sectors of approximately 10 m in diameter. This geometry was chosen to match the middle station layout. The detector modules, services and shielding of the NSWs must ﬁt in the 1110 mm thick envelope left by the current inner station. The NSW sectors combine the small-strip Thin Gap Chamber (sTGC) [7] and Mi- cromegas5 [8] technologies. Both detector technologies are read out with the VMM

5Micro-mesh gaseous structure

4 ASIC [9] which performs on-detector peak and time measurements of the detector signal. Individual sectors are assemblies of 2 Micromegas wedges placed between 2 sTGC wedges. Detector wedges are a combination of either 2 Micromegas modules or 3 sTGC modules. Different module types of varying size make up a wedge. The detector modules are quadruplets, meaning they have 4 independent detector layers. This multi-layer configuration was chosen for its robustness and redundancy. The NSWs are required to have an online angular resolution better than 1 mrad to match the resolution of the middle station required for the run following the Phase-II upgrade. In addition, the muon momentum resolution delivered by the NSWs must be comparable to that of the current inner station aiming for 15% for pT=1 TeV muons. Both requirements are satisfied with a spatial resolution better than 100 μm per detector plane. Finally, the NSWs must perform a bunch crossing identification of detector hits which requires, to be achieved, a time jitter better than 25 ns. A number of performance studies, some of which are described in the following, have proven the ability of the chosen NSW detector technologies to satisfy these specifications. Specificities of each detector technology are also described below.

4.1 sTGC technology Small-strip Thin Gap Chambers (sTGC) are multiwire chambers that operate in the quasi-saturated mode. The design of a sTGC is similar to that of the current muon spectrometer TGC detectors but with an improved spatial resolution and higher particle flux capabilities. The gas volume of a sTGC is contained between two segmented resistive cathodes. In the NSW configuration, each gas volume has one cathode segmented into strips and the other into a pad pattern. Strips are engraved with a pitch of 3.2 mm and are used for the precise measurement of the muon trajectory in the bending plane. Pads are used for the online selection of a specific band of strips to be read out after the passage of a muon thereby reducing the amount of strip data transmitted to the Level-1 trigger processor [10]. The area of pad electrodes varies between approximately 10 and 500 cm2. The time jitter of pads was measured in a beam test at CERN in 2012. The measured time distribution, shown in Fig. 3(a), demonstrates that more than 80% of the hits are within a time window of 25 ns [5]. The strip spatial resolution and the differential non-linearity bias were measured in a beam test at Fermilab in 2014 [11]. A spatial resolution better than 50 μm was measured with a perpendicular beam and obtained from the distribution of position difference between a reference pixel track and the sTGC measurement shown in Fig. 3(b). The charge sharing between neighbouring pads was measured in a beam test at CERN 2014 [11]. The measured charge asymmetry, as shown in Fig. 3(c), demonstrates that charge sharing occurs within a band of 5 mm, which is small compared to the overall dimensions of the pads.

5 A. Abusleme et al. / Nuclear Instruments and Methods in Physics Research A 817 (2016) 85–92 89

Table 1 200 Amplitude parameter ai for the differential non-linearity correction for three sTGC strip-cluster multiplicities. 150

Strip-cluster multiplicity i Amplitude parameter ai m] 100 µ

3 205 9 [ y) 7 δ 50 4 206 7 4 +

52117 5 pix 0

- (y' - −50

dimensional distributions in Fig. 5 (top). It shows the y-residual sTGC, 0 −100 y versus strip-cluster position relative to the closest inter-strip gap rel −150 centre ysTGC; 0. This effect is corrected using a sinusoidal function according to: −200 −5 −4 −3 −2 −1 012345 rel ysTGC ysTGC; 0 ai sin 2π ysTGC; 0 1 ¼ À ð Þ x'pix [mm] where ysTGC; 0 is the strip-cluster mean resulting from the Gaussian fi 200 t and ysTGC is the corrected particle position estimator. The amplitude parameters are denoted ai for the 3, 4 and 5 strip- 150 92 A. Abusleme et al. / Nuclear Instruments and Methods in Physics Research A 817 (2016) 85–92 multiplicity categories; the index i denotes the corresponding 100 category. These amplitude parameters are free parameters in the 7. Conclusions fi fi m] 12cmx12cm scintillator Beam

t. The values of the amplitude parameters obtained from the t to µ

[ 50 data are compatible with being equal for the three strip-cluster coincidence pix The spatial resolution for a full-size sTGC prototype detector for multiplicities as shown in Table 1. The correction function is 0 - y the ATLAS NSW upgrade has been measured in a 32 GeV pion therefore universal and is shown in Fig. 5 (top). The two- beam test experiment at Fermilab. A six-layer silicon pixel tele- sTGC −50

dimensional distribution after the correction is applied was y scope has been employed to characterize the sTGC detector and to found to be reasonably flat as shown in Fig. 5 (bottom). −100 correct for differential non-linearity of the reconstructed sTGC The alignment of the coordinate system of the pixel telescope strip-cluster position. At perpendicular incidence angle, single with respect to the above-defined coordinate system of the sTGC −150 strip-layer position resolutions of better than 50 m have been layer also affects the measured residual distribution. A simple two- μ −200 obtained, uniform along the sTGC strip and perpendicular wire parameter model is used to account for translations and rotations −5 −4 −3 −2 −1 012345 of the two coordinate systems with respect to each other. Both the directions, well within design requirements. The characteristics of Integration studies of the final VMM prototypex' [mm] with sTGC detectors are ongoing at readout region alignment correction and the differential non-linearity correction pix layer strip First the sTGC pad readout have been measured in a 130 GeV muon beam test at CERN. The transition region between readout pads are included in situ in the analysis. The alignment correction is Fig. 6. Two-dimensional distribution of y-residual versus xpix0 for a representative CERN.introduced in the model by expressing the pixel track position in data taking run and sTGC strip-layer. For the top figure the rotation alignmentFig. 14. Schematics of the experimental setup for charge sharing measurements. has been found to be 4 mm, and the pads have been found to be correction (red line) has been omitted when computing y-residuals, whereasThe beamthe and the scintillator coincidence area cover the transition between pad n fully efficient. the sTGC-layer coordinate system ypix, as a function of the track bottom figure shows the corrected y-residual distribution. and pad n 1. position in the pixel telescope coordinate system xpix0 and ypix0 , and þ two misalignment parameters δy and ϕxy, as follows: 1 Acknowledgements ypix 350xpix0 sin ϕxy ypix0 cos ϕxy δy: 2 ¼À þ þ ð Þ 600 σ = 43.4 ± 0.8 µm Run A, layer 4 The variable δy corresponds to a misalignment along the y-axis of The authors would like to thank the members of the NSW the sTGC300 coordinate system, and ϕ corresponds to a rotation of xy 2.85 kV 500 collaboration for their contributions and in particular the NSW the telescope coordinate system in the x–y plane around the z-axis 0.5 electronics group for providing the VMM1 based readout system. 250 Data of the sTGC coordinate system. Translation and rotation mis- We would also like to thank B. Iankovski, F. Balahsan, G. Cohen, S. alignments along and around the other axesSimulation are not taken into 400 Sbistalnik, A. Klier from the Weizmann Institute of Science, M. B. account200 in this model, since they are expected to have a small Moshe from Tel-Aviv University, M. Batygov, M. Bowcock, Y. Bar- impact on the determination of the intrinsic position resolution. 0 F ibeau, P. Gravelle from Carleton University, R. Openshaw from Fig. 6 shows the two-dimensional distribution of y-residual versus 300

150 Events TRIUMF and J. Fried from BNL for supporting this work. We are x pix0 for a representative data-taking period (run) and sTGC strip- fi fi also thankful to the rm MDT SRL (Milan) for their efforts in layer.Entries/0.78ns In the top gure the rotation alignment correction has been 200 producing the first prototypes of large cathode boards, which omitted100 when computing y-residuals. The mean of the residual -0.5 made the construction of the tested module possible. We distribution increases linearly as a function of x0 , which is evi- pix acknowledge the support of of the Israeli Science Foundation (ISF), dence50 for a small rotation25ns between the two coordinate systems. 100 The red line represents the correction applied to this dataset. CIRP-Israel-China Collaboration Grant 549/13, the MINERVA – Accounting0 for this correction results in a distribution that is foundation project 711143, the Benoziyo Center for Particle 340 350 360 370 380 390 400 410 420 430 440 0 -1 Physics, the Natural Sciences and the Engineering Research independent of xpix0 as shown in Fig. 6 (bottom). −400 −300 −200 −100 0100200300400 fi -8 -6 -4 -2 0 2 4 6 8 Council (NSERC) of Canada and CONICYT, Chile. The global model tted to theTime data (ns) is the following double y - y [µm] Gaussian function: sTGC pix First strip layer cluster position [mm]

rel Fig. 7. The residual distribution after all corrections are applied together withFig. 15.the Fraction of the charge collected by pad n as a function of the position with Fi Fi ysTGC; 0; ysTGC; 0; xpix0 ; ypix0 ; δy; ϕxy; ai; σ; f ; σw ¼ ð Þ result for the intrinsic resolution parameter σ for Run A, strip-layer 4. respect to the centre of the transition region. F 1 implies all the charge is References (a) (b) (c) ¼ fGysTGC ypix; 0; σ 1 f G ysTGC ypix; 0; σw ; deposited in pad n, while F -1 means all the charge goes to pad n 1. Figure 20: Simulated time¼ spectrumð À at a highÞþð À voltageÞ ð ofÀ 2.85 kVÞ compared with one measured in the ¼ þ [1] ATLAS Collaboration, Journal of Instrumentation 3 (2008) S08003. beam test with 180 GeV muons using 0.78 ns resolution TDC modules. The muonnarrow tracks Gaussian are perpendic- and it is typically around 95% with a RMS of [2] ATLAS Collaboration, ATLAS Muon Spectrometer Technical Design Report, fi 6.2. Charge sharing between pads ular to the detection planewhere andG denotes the horizontal a Gaussian axis function. of the The plot parameter has anf arbitrarydetermines o↵set.about 2%. The rst Gaussian represents the core of the residual CERN/LHCC 97–22, 1997. Figurethe relative 3: normalization (a) Measured of these two Gaussian pad functions. time The distribution. distribution The width parameter comparedσ is the parameter to of interest results from simulation [5]. [3] ATLAS Collaboration, New Small Wheel Technical Design Report, CERN-LHCC- value of f represents the fraction of the data parameterized by the for the determination of the intrinsic position resolution. To study the transition region between pads, the scintillator 2013-006, 2013. coincidence triggering area and the particle beam were centred [4] HL-LHC Collaboration, The High Luminosity LHC Project, 〈http://hilumilhc.web. (b) Difference between sTGC position measurements andbetween a precision pad n and pad n pixel1 of the first reference layer, as illustrated in cern.ch/about/hl-lhc-project〉. 6 Strip charge sharing þ [5] S. Majewski, et al., Nuclear Instruments and Methods in Physics Reasearch A Fig. 14. 217 (1983) 265. track. The sTGC intrinsic spatial resolution σ correspondsAfter to applying the timing width quality requirements of the on the dis-strip and pad [6] G. De Geronimo, et al., IEEE Transactions of the Nuclear Science NS-60 (3) Avalanches initiated by the ionized electrons and developed near wire surfaces will induce charges on hits, the channel baseline values are subtracted from the analog (2012) 2314. [7] V. Smakhtin, G. Mikenberg, A. Klier, et al., Nuclear Instruments and Methods in the two symmetrictribution cathode planes. The [11]. charges (c) spread Charge on the cathode asymmetry plane, however, are constrained between neighbourpeak sTGC values. Strip-clusters pads with[11]. induced charge in either 3, 4 or Physics Research A 598 (2009) 196. to be similar in size to the gas gap. Hence the charge sharing on 3.2 mm pitch readout strips underneath 5 adjacent strips are selected and calibrated in the same way as for [8] G. Mikenberg, D. Milstein, V. Smakhtin, et al., Nuclear Instruments and the Fermilab beam test. Events with a single strip-cluster in the Methods A 628 (2011) 177. is insucient to determine the charged particle hit position using a charge interpolation method. On fi [9] G. Mikenberg, D. Milstein, V. Smakhtin, et al., Test of special resolution and rst layer and the second layer are selected. The strip-cluster trigger efficiency of a combined Thin Gap and Fast Drift Tube Chambers for the other hand, the localized charge on the cathode will di↵use through the resistive paint layer to the position (mean of the fitted Gaussian) in the first layer is used to high luminosity LHC upgrades, 2011 IEEE Nuclear Science Symposium Con- define the position of the particle going through the detector. The ference Record NP5.S-200. ground and image charges will be created on the readout strips, which are capacitively coupled to the [10] I. Rubinskiy, An EUDET/AIDA pixel beam telescope for detector development, events are further required to contain a hit above threshold on resistive cathode. Therefore the resistive cathode not only provides the spark resistant functionality but Proceedings of the 2nd International Conference on Technology and Instru- either pad n or pad n 1. The charge fraction (F) is defined using mentation in Particle Physics (TIPP 2011), vol. 37, 2012, pp. 923–931. þ also serves to disperse4.2 the charge Micromegas over several readout strips to enhance technology tracking resolution. the analog peak values (P) of the two adjacent pads: [11] Arduino, UNO, 〈https://www.arduino.cc/en/Main/arduinoBoardUno〉. [12] C. Hu-Guo, et al., Nuclear Instruments and Methods in Physics Research A 623 A simple model has been developed to describe the charge sharing between sTGC strips. The model Pn Pn 1 (2010) 480. F À þ 3 considers the spatial distribution and time development of raw induced charges on the cathode, and ¼ Pn Pn 1 ð Þ þ þ the propagation ofMicromegas induced signals through are the resistive-capacitive micro-pattern network formed gaseous between the detectors restive with a thickFig. 15 shows 5 mm the charge drift fraction as gap a function and of the positiona cathode and the readout plane. with respect to the centre of the transition region between the Information aboutthin the raw 128 inducedμm charge amplification on the cathode can be obtained gap from separated Garfield simulations by a micro-meshpads. transparent It shows that the transition toregion, electrons. where the two pads share more than 70% of the induced charge, spans about 4 mm. without taking resistive layer into account. The charge dispersion phenomenon in resistive electrodes has been studied fromMicromegas streamer tubes in 1980’s operate [21, 22] to micro with pattern gaseous a moderate detectors [23, 24] electric recently. field of 600 V/cm in the drift region and The method developeda strong is also applicable electric for sTGC. field of 40 to 50 kV/cm in the amplification region. The ionization 6.1 Raw chargeelectrons spread on cathode created by a muon traversing the drift gap move towards the micro-mesh, For multi-wire chambers,and the eventually induced charge density reach on a contiguous the high-fieldcathode plane due to region a point charge where electron multiplication takes place. deposited on the wire is described by the empirical function of Gatti et al. [25, 26]: The detector signal is picked up by 300 μm wide copper readout strips located opposite 1 tanh2K = 2 ( ) K1 2 (8) to the micro-mesh1 + K3tanh andK2 etched on a readout PCB with a pitch of 415 μm. A layer of resistive strips glued over a thin kapton layer is installed above the readout strips aiming to improve the stability of the electric field in the amplification region and 17 reduce the spark probability. Micromegas quadruplets have two layers equipped with azimuthal strips, called η-strips, and two layers equipped with strips arranged in a stereo-strip configuration at an angle of 1.5◦ with respect to the η-strips. This strip configuration allows for a precision muon± position measurement in the bending plane and a coarse measurement in the coordinate perpendicular to the bending plane. The intrinsic spatial resolution of Micromegas detectors has been measured in several beam test campaigns from 2012 to the present, first using prototypes of small dimensions and then on full-size modules. The spatial resolution as a function of the track incidence angle obtained for small dimension prototypes is shown in

6 Fig. 4(a) [12]. The spatial resolution obtained when taking the strip charge cluster centroid is better than 90 μm but degrades as a function of the track angle. The resolution obtained using the μTPC mode [12], which uses the timing of strip hits, improves as a function of the angle of incidence. The time distribution of the ﬁrst strip with a hit, shown in Fig. 4(b) was also measured as a way to characterize the online timing performance [5]. Most hits are within a time window of 75 ns.

700 m]

µ µTPC mode 600 Centroid Combined 500 resolution [ 400

300

200 2014 JINST 9 C01017 100

0 10 15 20 25 30 35 40 angle [deg]

Figure 9. MicroMegas spatial resolution(a) summary as a function of the track angle. (b) Figure 5.9: Left: comparison of data to Monte Carlo of the earliest arrival signal time of a hit with the Figure 4: (a) Intrinsic spatial resolution of a Micromegasfirst generation as a function of the VMM of the readout particle chip. Right: Difference between the first arrival strip as track angle using different analysis techniques [12].obtained (b) Time from distribution offline from the of the read first out of the VMM1 chip and the strip selected from the strip hit compared to results from a Monte CarloVMM1 simulation for the [5]. trigger. Both Red results histogram have refers to teast beam data, blue histogram to Monte been obtained with small dimension prototypes. Carlo expectation.

Front-end readout electronics was based on the 128 channels APV25 ASIC in which detector signals are zero-suppressed, shaped with a CR-RC circuit and sampled at 25 ns frequency. The 5 Detector manufacturingtotal acquisition time window was of 675ns, corresponding to 27 samples.

Each NSW combines a total of 96 sTGC5.4.1 and Spatial 64 Micromegas resolution quadruplets for straight that and are inclined tracks. The µTPC method Figuremanufactured 10. Garfield simulations [7 in] for memberthe total drift velocity institutes (left) and the expected of the Lorentz ATLAS angle (right) collaboration. The production in Ar:CO2 93:7 gas mixture for a magnetic field up to 2 T as a function of the driftThe electric spatial field, with the resolution of MM detectors was studied for eight resistive chambers (T1–T8), aligned electricof and detector magnetic field vectors components being perpendicular. such as the sTGC cathode boards and Micromegas readout along the beam line and oriented, in pairs, in a back-to-back configuration. The total lever arm boards is carried out in collaboration with industry. The final detector assembly and 3 Performance of the MicroMegas in magnetic field of the system was 600 mm. The detectors were operated with an Ar:CO2 gas mixture (93:7). testing are realized at CERN. The reference operating settings were 600 V/cm electric drift field and an amplification voltage The MM chambersThe NSW of the NSW performs will operate in a a magnetic precise field reconstructionwith large variations and values of muon track segments which calls up to about 0.3 T, with different orientations with respect to the chamber planesHVmesh and a=500 sizeable V. componentfor a orthogonal good to control the MM electric of field. construction non-conformitiesData with beam during perpendicular detector manufacturing.to the MM and at various angles have been recorded and analyzed. InThe effect particular, of the magnetic the field on position the detector operation of each has been strip studied mustwith test beam be data known with an accuracy of 40 μm and simulations. Figure 10 shows the Garfield simulations [7] for the drift velocityThe and trigger the Lorentz was provided by the coincidence of three scintillators plus a veto. The APV25 chips anglealong as a function the of theprecision drift electric field coordinate for several values and of the80 magneticμmwere field along (perpendicular operated the beam at a frequency to achieve of 40the MHz, target completely NSW unsynchronized to the particle beam. A jitter to electricmomentum field) and for an resolution. Ar:CO2 93:7 gas mixture. Stringent Figure 11 illustrates tolerances the effectof of12.5 on the magnetic the ns, geometrycorresponding of the to the readout width strips of one are clock cycle, is introduced by the synchronization of ± enforced during construction. The alignmentthe trigger between signal. individual strip boards of a quadruplet is achieved using precision pins.The Many spatial steps resolution of the of assembly MM with process sub-mm are strip pitch and analog readout can easily go below –8– carried out on a granite table to ensure100 a goodµm for planarity perpendicular of the detector tracks by planes. using the cluster charge centroid method [31]. A spatial This section describes specificities ofresolution the sTGC of about and Micromegas 73 µm has been manufacturing. obtained for all chambers under test, with an average cluster The status of detector production at thesize time of approximately of writing is given. 2.4 strips. With the same perpendicular tracks, global detector inefficiencies have been measured to be in the range of 1–2%, consistent with the partially dead area expected from the presence of the 3007 µm diameter pillars separated by 2.5 mm. For impact angles greater than 10 the µTPC method [31] is used for a local track segment reconstruction in the few-millimeter wide drift gap. It exploits the measurement of the hits time and the highly segmented readout electrodes: the position of each strip gives an x coordinate, while the z coordinate (perpendicular to the strip plane) can be reconstructed from the time measurement of the hit after calibrating the z–t relation (z = t v ), see Fig. 5.10, left. ⇥ drift The hit time is measured for each strip by applying a fit to samples obtained from the shaped

56 5.1 sTGC manufacturing The sTGC production is divided in 5 productions lines, each responsible for manufacturing one or two quadruplet types. In some cases, different steps of the fabrication and quality control process of a production line are shared among different institutes6. Series production of sTGC modules is well ongoing in the production lines: several quadruplets have been manufactured and received at CERN. The assembly of the first sTGC wedge will be finished during Fall 2018. The production of cathode boards and other detector components is done in parallel to quadruplet assembly. All sTGC cathode boards will be produced by the end of 2018.

5.2 Micromegas manufacturing A total of 5 institutes are in charge of Micromegas quadruplet production7. Each institute is responsible for the manufacturing of one quadruplet type. As for the sTGC production, the integration of Micromegas quadruplets to form NSW wedges is done at CERN. At this stage, 70% of the required readout boards have been manufactured and production will be completed at the beginning of 2019. Series production of drift and readout panels, the main components of Micromegas quadruplets, is ongoing. The assembly of quadruplet is started in all construction sites. First quadruplets have been received at CERN and wedge integration is on the way to start.

6 Summary

The LHC instantaneous luminosity will increase by up to 5 to 7 times the design value following a series of upgrades planned over the next decade. In order to benefit from the physics opportunities offered by an upgraded LHC, the New Small Wheels (NSWs) will replace part of the muon end-cap station of ATLAS as a mean to improve the online muon identification capability. The NSWs will perform the precise reconstruction of candidate muon track segments and will combine the sTGC and Micromegas detector technologies. Detector construction for the NSWs is a worldwide effort shared among multiple physics institutes. Module manufacturing is well ongoing and wedge assembly has started.

6The sTGC construction sites are PUC (Chile), Shandong (China), TRIUMF, Carleton and McGill (Canada), Technion, TAU and Weizmann (Israel), and PNPI (Russia). 7The Micromegas construction sites are INFN (Italy), BMBF (Germany), Paris-Saclay (France), JINR (Russia) and Thessaloniki (Greece).

8 References

[1] L. Evans and P. Bryant (editors), LHC machine, JINST 3 (2008) S08001. [2] G. Apollinari et al., High-Luminosity Large Hadron Collider (HL-LHC): Technical Design Report V. 0.1, CERN-2017-007-M, CERN, Geneva, 2017, URL: https://cds.cern.ch/record/2284929. [3] ATLAS Collaboration, The ATLAS Experiment at the CERN Large Hadron Collider, JINST 3 (2008) S08003. [4] P. Vankov on behalf of the ATLAS Collaboration, ATLAS Future Upgrade, PoS 273 (2016) 061 proceedings of BEAUTY2016, ATL-UPGRADE-PROC-2016-003. [5] ATLAS Collaboration, New Small Wheel Technical Design Report, CERN-LHCC-2013-006, CERN, Geneva, Jun, 2013, URL: https://cds.cern.ch/record/1552862. [6] ATLAS Collaboration, Performance of the ATLAS muon trigger in pp collisions at √s = 8 TeV, Eur. Phys. J. C 75 (2015) 120. [7] S. Majewski et al., A thin multiwire chamber operating in the high multiplication mode, Nucl. Instrum. Methods Phys. Res. 217 (1983) 265. [8] Y. Giomataris et al., Micromegas: a high-granularity position-sensitive gaseous detector for high particle-ﬂux environments, Nucl. Instrum. Meth. A 376 (1996) 29. [9] G. Iakovidis on behalf of the ATLAS Muon Collaboration, VMM - An ASIC for Micropattern Detectors, EPJ Web Conf. 174 (2018) 07001 ATL-MUON-PROC-2015-015. [10] L. Guan on behalf of the ATLAS Muon Collaboration, Trigger algorithms and electronics for the ATLAS muon new small wheel upgrade, JINST 11 (2016) C01083. [11] A. Abusleme et al., Performance of a full-size small-strip thin gap chamber prototype for the ATLAS new small wheel muon upgrade, Nucl. Instrum. Meth. A 817 (2016) 85. [12] M. Iodice on behalf of the Muon ATLAS MicroMegas Activity (MAMMA) Collaboration, Performance studies of MicroMegas for the ATLAS experiment, JINST 9 (2014) C01017.